Modelling of Heterogeneously Distributed Soluble Boron and Coolant Density Within a Fuel
Assembly
BJÖRN EKLÖF
Master of Science Thesis Division of Reactor Physics, KTH
Supervisors: Janne Wallenius Marcus Eriksson Examiner: Janne Wallenius
TRITA-FYS 2011:64
iii
Abstract
This thesis will investigate the implications of extending a cross-section model based on flat, homogeneous, boron distribution and density states within a fuel assembly, with correction terms from heterogeneously distributed boron and coolant density. The heterogeneous states refers to differences in active coolant and bypass coolant states (which is not in contact with the fuel). The aim is to verify the implementation of this model in the simulator software POLCA-T, used at Westinghouse to simulate BWR transients, as well as to verify the model for use in transient simulations. This was performed by solving differ- ent transient problems and by comparing the results with other cross-section models.
The standard way of modeling the bypass is to average all bypass sections into one collective, averaged bypass. In this thesis, the effect of modeling with mul- tiple bypass channels, and the implications on
heterogeneities, have been investigated.
It was found that in many cases, differences in coolant flow velocities be- tween the active regions and bypass region generate heterogeneous boron states.
These heterogeneities become large enough for the heterogeneous effects on re- activity to be observed. It was also found that in the event of loss of pressure and onset of bypass boiling, heterogeneous density states with noticeable re- activity effects were present. The investigation into multiple bypasses lead to inconclusive results.
The evaluation confirms that the model has been correctly coupled to POLCA-T and performs as intended.
Acknowledgements
I would like to thank my mentor Marcus Eriksson for his support and invaluable feedback. For his time spent on reading and ensuring a high academic standard on this report, I would like to thank Janne Wallenius at the Division of Reactor Physics at KTH. I would also like to thank Westinghouse Electric Sweden AB in Västerås for allowing me access to their extensive knowledge, industry leading technology and time from their employees, especially Ulf Bredolt who helped me with a lot of bug fixes and implementing new features as needed, and Yonatan Dag for preparational work on the subject.
v
Contents
List of Figures ix
List of Tables xi
1 Introduction 1
1.1 Aims and Objectives . . . . 2
1.2 Terminology . . . . 2
1.3 Reactor Model . . . . 2
1.3.1 Bypass . . . . 3
1.4 Boron Deployment in Nuclear Reactors . . . . 4
1.5 Void Interaction . . . . 6
2 Methodology 9 2.1 The Simulator, POLCA-T . . . . 9
2.1.1 Power Generation Model in POLCA7 . . . . 10
2.1.2 Cross sections in POLCA7 . . . . 10
2.2 Multi Group Theory . . . . 11
2.3 Models Descriptions . . . . 12
2.3.1 Standard POLCA7 cross section Model . . . . 12
2.3.2 Effective Coolant Density Model . . . . 13
2.3.3 Effective bypass-boron Model . . . . 13
2.3.4 Heterogeneous Model . . . . 14
2.4 The Heterogeneous Model in Depth . . . . 14
2.4.1 Aim & Purpose of the Model . . . . 14
2.4.2 Void Contribution . . . . 15
2.4.3 Boron Contribution . . . . 15
2.4.4 Cross-Correlation Contribution . . . . 16
vii
3.2 Full Scale ABWR ATWS . . . . 34
4 Discussion of Results 37 4.1 General . . . . 37
4.2 Single bypass . . . . 39
4.3 Multiple bypasses . . . . 40
4.4 Realistic ATWS Simulation . . . . 42
4.5 Uncertainties in Results . . . . 42
5 Conclusions 45 5.1 Suggestions For Further Work . . . . 46
A Appendix 47 A.1 Fuel cost . . . . 47
Bibliography 49
List of Figures
1.1 Thermal power profile of quarter symmetry ABWR core. . . . . 3
1.2 Topological map of a BWR fuel bundle . . . . 4
1.3 Neutronic cross section for b
10. . . . 5
1.4 Neutronic cross section for water . . . . 6
1.5 Neutron cross section data for U
235. . . . 7
2.1 Heterogeneous cross term for current- and historic densities. . . . 18
2.2 Heterogeneous boron term for current- and historic densities. . . . 18
2.3 Heterogeneous density term for current- and historic densities. . . . . . 19
2.4 Overview of reduced model. . . . 20
2.5 Bypass mappings of quarter symmetry nuclear reactor core . . . . 22
3.1 Case 1, 300 kg/s boron injected, k
effvs time. . . . 26
3.2 Case 1, 300 kg/s boron injected, boron concentration & concentration difference vs. time, from central active bundle and bypass. . . . 26
3.3 Case 2, circulation flow reduced to 1000 kg/s, k
effvs time. . . . 27
3.4 Case 2, circulation flow reduced to 1000 kg/s, bypass density difference from reference vs. time. . . . 27
3.5 Case 3, 7 kg/s boron injected, circulation flow reduced to 1000 kg/s, k
effvs time. . . . 28
3.6 Case 3, 7 kg/s boron injected, circulation flow reduced to 1000 kg/s, boron concentration & concentration difference vs. time, from central active bundle and bypass. . . . 28
3.7 Case 3, 7 kg/s boron injected, circulation flow reduced to 1000 kg/s, bypass density difference from reference vs. time. . . . 29
3.8 Case 4, 300 kg/s boron slowly injected, k
effvs time. Long runtime. . . . 29
ix
equal size bypass. . . . 33
3.17 Case 6, circulation flow reduced to 1000 kg/s, ρ
bpvs. time. 5-channel bypass. . . . . 34
3.18 ATWS power comparison . . . . 34
3.19 ATWS k
effcomparison . . . . 35
3.20 ATWS cladding temperature comparison . . . . 35
3.21 ATWS containment temperature rise comparison . . . . 36
3.22 ATWS boron distribution . . . . 36
List of Tables
A.1 Calculation of fuel costs for 1 kgU enriched fuel in units of thousand US$. 47
xi
Chapter 1
Introduction
The importance of simulators and their applications in science and engineering have grown substantially over the last decades. Driven mainly by hardware improve- ments, faster computers have made more complex simulations possible, relying less and less on simplifications and assumptions on symmetry for acceptable run times.
In the context of nuclear engineering simulators have been used for a long time, mainly to predict fuel burn and to optimize fuel economy, but also for safety anal- ysis. Fuel optimization is a matter of improving cost efficiency and has historically driven development towards more complex software. With fuel costs currently at
∼ e100 per kg UF
6[1], resulting in an end price of around e1800 per kg enriched Uranium (see appendix A.1); the incentive to maximize burnup is more important than ever. By taking storage and handling of the waste products into considera- tion, the incentive to minimize volume becomes even stronger. In later years safety concerns have risen in importance, wherein transient analysis plays a central role as malfunctions give rise to chains of events which are intrinsically transient, and so the analysis must go beyond steady state.
This thesis will investigate the effects and properties of implementing a model for heterogeneously distributed boron and density in the context of transient simula- tions in Westinghouse’s transient simulation code POLCA-T.
1.1 Aims and Objectives
The aim is to verify the coupling of the heterogeneous cross-section model to POLCA-T. Analysis is to be performed to verify that the model gives results
1
eff
neutrons)/#(absorbed + leaked neutrons). A value greater than one signifies rising neutron density and increasing number of fissions per second, i.e. a supercritical core. A value less than one signifies decreasing neutron density and fewer fissions per second, i.e. a sub-critical core. A value of exactly one means an equilibrium core, with the same number of neutrons produced as used.
Reactivity coefficients:
Reactivity is a macroscopic measurement, which can be analyzed on a local or global scale. The reactivity depends on the macroscopic cross section, which is made up by reactivity coefficients contributing positive or negative reactivity depending on whether the phenomena described increases or decreases the macroscopic cross sec- tion. The reactivity coefficients thus are related to the cross section models and are by no means physical constants.
1.3 Reactor Model
The reactor simulated is the Westinghouse Advanced Boiling Water Reactor (ABWR), an internal recirculation pumps BWR with a rating of 3,93 GW
thdeployed with SVEA Optima-2 fuel loaded as an equilibrium core. The reactor core has 872 fuel bundles and 205 control rods. The nominal specific power density is 25,3 kW/kgU.
The simulations carried out in this report are at the end of the fuel cycle with an average exposure of 37,0 MWd/kgU.
1.3.1 Bypass
The fuel assembly simulated is the SVEA-96 Optima2 fuel. It consists of four cells
containing 24 fuel pins each, arranged on a square. The coolant in direct contact
with the fuel pins, inside the cells, is defined as the active coolant. The flow paths
between the cells, as well as between the assemblies, are defined as bypass regions.
1.4. BORON DEPLOYMENT IN NUCLEAR REACTORS 3
Figure 1.1: Thermal power profile of quarter symmetry ABWR core.
Within each assembly there is one bypass water cross (WX) in the middle and four water wings, separating the cells. These five bypass regions are collectively called the intra-assembly bypass. Between the assemblies lays the inter-assembly bypass.
The bypasses are not in direct contact with the fuel and act as a neutron moderator and thermal buffer.
When simulated, all bypass regions for each assembly are added to a lumped assem- bly bypass, marked red in figure below. The standard method is then to average all lumped assembly bypasses into one effective bypass channel. An investigation into mapping the assembly bypasses into several channels laid out in radial regions has been performed.
1.4 Boron Deployment in Nuclear Reactors
The boron cross section mentioned in the section above is for the case of emergency shut down of the nuclear reactions. If the control rods can’t deploy, boron tanks are present to pump borated water directly to the upper plenum, right above the core. The system can either be manual or automatic, triggered by sensors and event chains.
The reason for boron being used for this purpose is due to its large cross section,
and thus its ability to absorb thermal neutrons efficiently and making the reactor
Figure 1.2: Svea-96 Optima2 fuel assembly. Bypass regions marked red.
sub-critical. The isotope B
10has a high cross section, almost 10
4barns for thermal neutron (see figure 1.3), compared to the naturally more abundant B
11, with cross sections of about 5 barns for thermal neutrons [9].
The nuclear process of neutron absorption for B
10[4]:
B
10+ n −→ B
11∗Li
7+ He
4+ γ + 2,4 MeV (1.1) In typical cases of boron injection the concentrations reaches a couple of hundred ppm’s (relative weight), which equates to number densities
1of ∼ 10
27m
−3. This can be compared to the typical neutron number density
2of ∼ 10
19m
−3, which ensures that there is more than enough boron present to stop reactions and no risk for the boron to burn out.
The fact that water has a cross section of about 50 barn compared to 10
4barn for B
10for thermal neutrons, seen in figure 1.3 & figure 1.4 (together with number densities of 6,24 · 10
30m
−3for water and 10
27m
−3for B
10), ensures that the boron
1Concentration is defined as mass relatives, but since the mass difference between boron and water is ∼ 2 and only order of magnitudes are compared, with saturated liquid water density under operation: ∼ 740 kg/m3⇒ 6,24 · 1030m−3.
2power = fissions/s · energy/fission ⇒ fissions/s = power/(energy/fission). Fissions/s ' 4 GWth/200 MeV = 2,5·1028/2·108s−1= 1,14·1020s−1⇒ total number neutrons (at equilibrium)
= 2,5 · 1,14 · 1020≈ 3 · 1020. The neutron number density then becomes (with nuclear core volume of ∼ 64 m3): ∼ 5 · 1019 n/m3.
1.5. VOID INTERACTION 5
has high chances of interaction with the neutrons; The quotient of the water and the boron products of number density and cross section gives a simplified relative probability for interaction with the neutrons, which in this case is ∼ 0,1 for boron, or about a tenth the probability to interact with water.
The numbers stated above are for a BWR (Boiling Water Reactor). In BWR’s natural boron is used, with an isotope distribution of 80,18 % B
11and 19,82 % B
10[2], and thus only about 20 % of the mass contributes significantly to the macro- scopic cross section. The situation is slightly different for PWR’s (Pressure Water Reactor), since they operate under much higher pressures they need higher concen- trations of active boron to reach adequate probabilities of interaction with thermal neutrons. To reach the higher concentrations, boron enriched with B
10is deployed instead of natural boron.
Figure 1.3: Neutronic cross-section for B
10[8].
1.5 Void Interaction
Void fraction, or percentage of coolant in gas phase, generally has a strong impact on
reactivity. Since almost all commercial nuclear reactors are thermal reactors where
the main absorption is of moderated neutrons with kinetic energy comparable to
the thermo-dynamically available energy (thermal neutron: E
kin' k
BT ' 0,025 eV
). Since neutrons produced from nuclear reactions typically have kinetic energies
of ∼ 2 MeV [7], they need to be moderated (slowed down) to thermal energies to
increase the absorption probability, according to figure 1.5. If the neutrons can’t
be slowed down, the reaction probability will be low and many neutrons will leak
out of the nuclear core. By escaping the core, the neutrons available to initiate
Figure 1.4: Neutronic cross section for liquid water and the sum of its components [6].
new fissions decreases and if the conditions persist the nuclear rector will become sub-critical.
When void levels increase, the general density of the coolant decreases. Since the
coolant acts as the moderator in light-water reactors, the general loss of mass density
means a general loss of number density. With less water molecules present, the
probability of a neutron to strike a molecule and become moderated decreases,
which in turn leads to a longer mean free paths and a higher probability of escaping
the nuclear core. This is one of the main safety features in BWR’s, since if there
is a problem with the coolant flow, this will lead to loss of pressure which gives a
higher void fraction, which in turn slows or shuts down the reactor. In PWR’s, it is
true to some extent that void interaction acts as a safety feature, in the sense that
in case of a coolant leak this will lead to voiding which will lower the reaction rate.
1.5. VOID INTERACTION 7
Figure 1.5: Neutron cross section data for U
235[5]
Chapter 2
Methodology
2.1 The Simulator, POLCA-T
POLCA-T is a coupled computer code for transient thermal-hydraulics and neutron kinetics analysis of BWR and can be used as a general tool for advanced simula- tion of single and two-phase flow systems including none condensable gases. The code consists of POLCA7, a steady state solver for a given operating point, which in turn is introduced to a discretizised time line by POLCA-T. Execution of the steady state solver is performed repeatedly for the operating point the system re- sides in, utilizing the output of the previous solution as the input for the next time step. The code has a full 3D model of the reactor core. The reactor pressure ves- sel, external pump loops, steam system, feed water system, ECC (Emergency Core Coolant) systems and steam relief system can be modeled to the desired details.
Control and safety systems are modeled using a code package called SAFIR, which is linked to the code. The SAFIR package is in production use together with the 1D BWR transient code BISON.
The application areas of the POLCA-T cover operational transient, stability, RIA (Reactivity-Initiated Accident), ATWS (Anticipated Transients Without Scram), ATWC (Anticipated Transients Without Control rods) and LOCA (Loss Of Coolant Accident).
The neutronics models in the code are the same as those in the static nuclear core analyzer POLCA7 with addition of proper kinetic terms for transient use. The thermal-hydraulic model has advanced physics with thermal non-equilibrium and full geometric flexibility.
The POLCA-T code is well adapted to analyze the type of scenarios with a number
9
3D two group kinetics model allowing for up to six delay neutron groups calculates the fission power generation in the core. Reactivity feedback is included for void fraction, moderator (coolant) temperature, fuel temperature, and reactor control rods. The power fraction deposited in the fuel and coolant can be specified versus time or as a function of coolant density.
The decay power generation is calculated by the sum of eleven fission product decay groups and the actinide decay of Uranium-239 and Neptunium-239 or by a user provided table as decay power versus time.
2.1.2 Cross sections in POLCA7
POLCA7 relies on homogenized (spatially smeared) two-group macroscopic and mi- croscopic cross sections, diffusion coefficients and discontinuity factors to solve the neutron diffusion equation. The generation of the nodal data and their tabulation as functions of local physical states is done outside of POLCA7 by means of two- dimensional (2D) lattice physics calculations. Nodal data is generated for individual fuel segment types which is uniquely defined by its lattice design (geometry, enrich- ment, burnable absorber, loading, e.t.c.). The nodal data is generated as a discrete set of cross sections as functions of state parameters. The cross section for a given state point is then interpolated linearly from the closest tabulated points.
2.2 Multi Group Theory
Cross sections dictates absorption probabilities, and thus is a governing factor of reactivity. In nuclear reactors the reactivity is important for determining criticality, power generation and general dynamics of the core. The size of the cross sections depends on several factors, incident particle energy being among the most impor- tant.
Figure 1.3 & 1.4 show examples of how the cross sections are dependent on incident
particle (in this case neutron) energy.
2.2. MULTI GROUP THEORY 11
In designing POLCA-T, the balance between precision and the work needed for generating cross sections was chosen to be two groups for active neutrons, meaning that particles (neutrons) belong to either a fast or a thermal energy group. Delayed neutrons is represented by six groups and decay power from eleven groups. For each group separate cross sections for all different materials and state point combinations have to be calculated.
Neutronics is handled by POLCA7 through representation (for each group) by num- ber densities as functions of position and time. In each time step, the neutronic solver calculates the change in number density for each group in each discrete point, from diffusion, absorption and flows between the groups (e.g. one way flow towards lower energy). The rate of change of neutrons, multiplied with the number density gives the neutron flux for each spatial point.
Absorption rate depends on the cross section of each material present, for each
energy group as well as the composition of materials at each point. Diffusion de-
pends on the neutron density and gradient in a given point. Group transitional flow
depends on transition probabilities of the materials present and neutron number
density of the high energy group at a given node. The mathematical formalism is
explained in detail in [3].
2.3.1 Standard POLCA7 cross section Model
The standard POLCA7 cross section model is based on a combination of a base cross section and deviation terms. The base cross sections are valid at base conditions, a predefined set of states, not including exposure (E), historic coolant density (ρ
h) or current active coolant density (ρ). Cross sections for different state points are then interpolated from the nearest set of base conditions. The cross sections (both macroscopic and microscopic) takes the form:
Σ = Σ
b(E,ρ,ρ
h) + ∆Σ
CR(E,ρ,ρ
h) + ∆Σ
SG(E,ρ) + d
dop(E,ρ)
p
T
f−
qT
fb
+ +b
B(E,ρ,ρ
h) [C
B− C
B] + X
Xe(E,ρ,ρ
h) [N
Xe− N
Xe(E,ρ
h)] +
+Σ
iσ
i(E,ρ,ρ
h,CR, T
f, N
Xe, C
B)[N
i− N
ib(E,ρ
h)]+
+ H.O.T. + Cross terms + Intra − nodal correction terms (2.1) Where the terms account for control rods (CR), spacer grids (SG), Doppler effect (dop), soluble boron concentration (C
B) and Xenon (Xe) as well as heavy nuclide isotopic corrections. T
fdenotes the node average fuel Doppler temperature and C
Bthe average soluble boron concentration in units of (ppm).
Cell data in general are computed as functions of exposure (E), historic coolant
density (ρ
h) and current active coolant density (ρ). During generation bypass wa-
ter regions are set to correspond to the active coolant liquid density, i.e. a flat,
homogeneous state. The boron coefficients are computed based on homogeneous
(or flat) distribution throughout the assembly. The governing concentration when
using the cell data is the boron concentration of the active coolant.
2.3. MODELS DESCRIPTIONS 13
Regarding functional dependence of the fuel Doppler effect
The square root fuel temperature dependence seen in equation 2.1 is not a fun- damental physical relation of the Doppler effect, but rather a coincidence. For the neutron spectrum in thermal LWR’s, the temperature dependence of the Doppler ef- fect coincides with the square root function. For most parts of the kinetic spectrum this is not the case however, instead the temperature dependence have to be curve- fitted to experimental data and in general one can only say that the temperature dependence is proportional to a fractional exponent of temperature.
2.3.2 Effective Coolant Density Model
This model uses the standard cross section model with the difference that an effec- tive coolant density is used to look up cross sections in the cell data tables. This means that the coolant density ρ is replaced by an effective density, ρ
eff. The effec- tive coolant density model will conserve the total coolant mass in the active region plus the bypass region. The effective density applied in the cell data interpolations takes into account the deviation in the calculated bypass density from the reference bypass density used during cross section data generation. By default the ρ
refbpis set to ρ
liq,satbp, a reference density for liquid phase coolant. The model is derived from:
ρ
actA
act+ ρ
bpA
bp= ρ
effA
act+ ρ
refbpA
bp⇔ ρ
eff= ρ
act+ A
bpA
act(ρ
bp− ρ
refbp) (2.2) Where A
actis the active coolant flow area, A
bpis the bypass flow area, ρ
actis the current active coolant density, ρ
bpis the bypass density.
2.3.3 Effective bypass-boron Model
This model uses the standard cross section model with the difference that an effective boron concentration is used to look up cross sections in the cell data tables. The boron concentration C
Bis replaced by an effective concentration C
Beff. The effective boron will conserve total boron concentration in the active region plus bypass region.
The effective density applied in the cell data interpolations takes into account the deviation in bypass boron concentration used during cross section data generation.
By default the bypass boron concentration is set to be identical to the active coolant
boron concentration.
+ steam phase, ρ
bpis the liquid + steam phase density in the bypass channel.
2.3.4 Heterogeneous Model
The heterogeneous model uses specifically generated cell data to account for het- erogeneous states. By doing so, assumptions are made that differences in bypass density or boron concentration states affects the cross sections the same way as the same difference in active coolant. Both effective models make this assumption by only scaling the deviations with the bypass fractional area.
As with any linear interpolation, the further from the generation point one gets, the more the error tends to grow. Therefore the heterogeneous model should provide more accurate results especially at states further away from generated points. The next section describes the heterogeneous model in depth.
2.4 The Heterogeneous Model in Depth
2.4.1 Aim & Purpose of the Model
The heterogeneous model aims to improve the accuracy of which the cross section modeling in POLCA-T. In the event of voiding in the bypass or injection of borated water, the conditions assumed during cell data generation are no longer valid. Since the bypass makes up about 1/3 of the coolant flow area of the fuel assembly, any slight changes to state here will have a noticeable impact on the total macroscopic cross section. In the context of transient simulations, the heterogeneous model is an attempt to treat reactivity feedback originating from coolant voiding in the bypass and/or presence of soluble boron in greater detail.
The contribution to the macroscopic cross section is given by the expression:
∆Σ
het(E, ρ
his, ρ) = b
bypB(E, ρ
his, ρ) ·
C
Bbyp− C
Bact+
2.4. THE HETEROGENEOUS MODEL IN DEPTH 15
+b
bypρ(E, ρ
his, ρ) ·
ρ
byp− ρ
act,refliq+
+ b
bypB,ρ(E, ρ
his, ρ) ·
C
Bbyp− C
Bactρ
byp− ρ
act,refliq(2.4) Where b
YXis the tabulated cross section data, E is burnup, ρ
hisis historic density in kg/m
3, ρ
act,refliqis the active channel reference density for saturated water, ρ
bypis bypass instantaneous density in kg/m
3and C
Bis boron concentration in ppm for active and bypass channels respectively.
2.4.2 Void Contribution
In the event of voiding in the bypass, there will be a change in the macroscopic cross section due to loss of moderating water molecules (due to lowered density).
It is currently assumed that by using cross section data generated without bypass void, conservative calculations are performed since this assumption leads to higher reactivity; if there are margins to acceptance criteria with this assumption, there will be even larger margins if void is present in the bypass. The inclusion of bypass void corrections to reactivity will give more accurate results and will increase un- derstanding of actual behavior under such conditions. From equation 2.4, it can be seen that the void correction will increase reactivity if the bypass has higher density than the reference value used in generating the standard cross section data, and decreased reactivity if the density is lower than the reference value. The correction therefore compensates the cross- section if density differs from reference, and the assumption made in the standard method no longer holds.
2.4.3 Boron Contribution
The standard method for simulating injection of borated water is to assume equal concentrations in the active fuel channels and its surrounding bypasses. Since the boron transport in POLCA-T is based on local mass-balance, the boron flux depends mainly on the flow rate of the coolant in each region. Unless the flow velocity of the coolant in liquid phase, as well as the boundary conditions, are equal for both the active channel and its bypass, concentration gradients will emerge over the active channel and its bypasses, i.e. a condition of heterogeneous boron distribution. Since the diffusion of boron is temporal, it is a dynamic, transient process. When the boron has diffused and reached a condition of equilibrium, the heterogeneous effect vanishes. Thus as long as there is no rapid change in boron concentration fed to the reactor core, a homogeneous boron concentration distribution will be present and the standard cross section model is accurate.
The effect of injecting borated water with the heterogeneous model is such that if
the boron reaches the bypass first, the boron will have higher efficiency compared
to the standard cross section model. The physical explanation being that under
the size of the coefficient is three orders of magnitude smaller than the first order corrections, at its largest value (which is seen in figures 2.1, 2.2 & 2.3), and thus will for most cases have a very small contribution, overwhelmed by the first order corrections. Any such contribution will be close to impossible to identify in any global state parameters such as k
effor the thermal power, therefore no investigation into this specific term will be performed in this report. An illustrative example is given below showing the needed magnitudes of the perturbations to make this second order term of significant size. Only thermal neutrons are compared in the example, since the values for fast neutrons are two orders of magnitude smaller for all coefficients, thus their contribution to a global parameter such as k
effwill be significantly smaller. Furthermore the magnitudes for the fast neutron coeffi- cients have the same distribution as the ones for thermal neutrons (i.e. 2 orders of magnitude at its largest), therefore any claims to the relative importance made for thermal neutrons should be possible also for fast neutrons.
A simple example highlighting this follows, with cell data as:
BDHSA2 = 4,63 · 10
−10, BHSGA2 = 6,83 · 10
−6, DHSGA2 = −6,29 · 10
−7Where BDHSA2 is the cross-correlation coefficient, BHSGA2 the boron coefficient and DHSGA2 the density coefficient for the heterogeneous part of the macroscopic thermal absorption cross section.
These data are from an active volume cell of channel 202 (i.e. an active chan- nel close to the center of the core) in the ABWR, with conditions ρ
h=ρ=460 kg/m
3and mean burn up of 40 MWd/kgU, which are typical values for active channels under normal operation (38 % void).
4,63 · 10
−10x
2≥ 7,51 · 10
−6x
2.4. THE HETEROGENEOUS MODEL IN DEPTH 17
Here, the variable has been defined in favor of the cross-term having more influ- ence. With this inequality, one can show for what value of the variable x the cross term exceeds a preset condition (where x is the state variables of interest: boron concentration difference and density deviation from reference value).
For the cross term to grow to at more than 10 % of the total correction term, we look at the inequality:
4,63 · 10
−107,51 · 10
−6x ≥ 0,1 ⇔ x ≥ 1,62 · 10
3Where the most favorable conditions for the cross term has been assumed (i.e.
concentrations and density differences of the same magnitude). The result shows that concentrations and density differences from reference of the magnitude of about 1000 ppm boron concentration difference, as well as about 1000 kg/m
3in density difference, need to be present at the same time to make the cross term contribute more than 10 % of the total correction. Such large concentration differences as well as density differences are not present, at the same time, under realistic conditions.
For the condition that the cross term should contribute at least 1 %, the same
calculations give that magnitudes of 100 for both variables are needed at the same
time. When boron is injected the fission reactions stop abruptly, meaning that
the void will decrease and thus situations where both conditions are present at the
same time are rare. This in addition to the conservative assumptions means that for
realistic values, the cross term only contributes less than 1 % of the total correction,
which in turn confirms the previous statement, that the term is too small to analyze
exclusively. As was discussed before the example, in a state where the cross-term
is most important, i.e. where BDHSA2 is largest compared to when BHSGA2 is
smallest (from figure 2.1 and 2.2), the above calculations have to be scaled with
about 0.5; i.e. magnitudes of 800 or 80 for 10 % and 1 % contribution respectively.
Figure 2.1: Values for BDHSA2 (Boron & density cross coefficient) for current- and historic density state parameters, with burnup of 40 MWd/kgU.
Figure 2.2: Values for BHSGA2 (Boron coefficient) for current- and historic density
state parameters, with burnup of 40 MWd/kgU.
2.5. SYSTEM MODEL 19
Figure 2.3: Values for DHSGA2 (Density coefficient) for current- and historic den- sity state parameters, with burnup of 40 MWd/kgU.
2.5 System Model
The model used for verification is the Westinghouse ABWR, described in section 1.3. For analytic purposes, modification to the predefined system model will be covered as well.
2.5.1 Reduced System Model
All the transients run for purpose of the model verification were performed with
a reduced version of the full ABWR, compromising only of the core and the two
closest upper and lower plenum sections. The boundary conditions chosen used
were a constant pressure boundary at the upper plenum, and a constant circulation
flow at the lower plenum, set to the full power steady state values of the full ABWR
model. Figure 2.4 shows the schematics (with three active channels visible) of the
reduced model. The reason for reducing the model is to make sure that effects noted
are not caused by feedback and/or disturbances from the external systems, as well
as for shorting the run times and lower the tendency for diverging solutions.
approach, one average bypass channel is treated. This bypass channel exchanges
Figure 2.4: Overview of reduced model. Three active channels showing (out of 218
for quarter symmetry). The bypass and the external bypass are also showing, as
well as the boron container near the top left.
2.6. TEST CASES 21
heat with all (active) fuel channels throughout the assembly walls and internal wall surfaces. Heat is also directly deposited in the bypass by neutron and gamma radiation. As part of the increased focus on the bypass with the heterogeneous model, an investigation into modeling multiple bypasses were performed. With the heterogeneous model, the cross sections are directly influence by heterogeneous states within the bypass, why it is important to investigate if using several bypasses changes the results; since heterogeneous states can be local and thus only affect local neutronics, this can have implications for the results.
The following bypass mapping configurations were studied:
1. Two internal bypasses mapped similar to the throttle map.
2. Two internal bypasses with almost equal flow areas.
3. Five internal bypasses with almost equal flow areas and with outermost region similar to throttle map.
Due to convergence issues, the bypass map visualized in figure 2.5b was chosen not to strictly follow the throttle zone. By doing this, a slight averaging occurs due to mixing of sources of two different flow velocities into one channel. This channel will see a mass flow which lies in-between the mass flows of the two sources by themselves, thus reducing the strain on the boundary region between the two bypass channels modeled. The chosen mappings were tested to investigate what influence the mapping has to the simulation results, and to verify that there are no scaling errors or other partitioning effects on the results. In all cases of multiple bypass channels, there are no interconnections between the bypass channels, meaning that no transversal flow, coolant water or heat, can occur except at the upper and lower plenum sections.
2.6 Test Cases
The four models presented in section 2.3 is the only difference between the results within each test case. All other parameters such as disturbance, initial state and boundary conditions are identical.
2.6.1 Disturbances
The cases run were the following disturbances, for a single bypass configuration:
1. Borated water injected to lower plenum with a flow of 300 kg/s. Start from 0
kg/s, then at 1 s ramping up reaching 300 kg/s at 2 s, continuing at 300 kg/s
for the rest of the run. Coolant flow kept at nominal value of 14500 kg/s.
(a) Throttle zones of reactor core. (b) Bypass map with two channels, mapped after throttle zones.
(c) Bypass map with two channels, approxi- mately equal flow area for both channels.
(d) Bypass map with five channels, approxi- mately equal flow area for all channels.
Figure 2.5: Bypass mappings of quarter symmetry nuclear reactor core, seen from
above.
2.6. TEST CASES 23
2. Coolant flow reduced to 1000 kg/s. Start from 14500 kg/s, then at 1 s ramping down to 1000 kg/s at 6 s, continuing at 1000 kg/s for the rest of the run.
3. Borated water injected to lower plenum with a flow of 7 kg/s. Start from 0 kg/s, then at 1 s ramping up reaching 300 kg/s at 2 s, continuing at 7 kg/s for the rest of the run. Coolant flow reduced to 1000 kg/s. Start from 14500 kg/s, then at 1 s ramping down to 1000 kg/s at 6 s, continuing at 1000 kg/s for the rest of the run.
4. Identical to case 1, but longer run. Borated water injected to lower plenum with a flow of 300 kg/s. Start from 0 kg/s, then at 1 s ramping up reaching 300 kg/s at 300 s, which is end of run. Coolant flow kept at nominal value of 14500 kg/s.
For multiple bypasses (2 bypass channels mapped after throttle zones, 2 bypass channels with equal flow area and 5 bypass channels with equal flow area) all runs have the heterogeneity model enabled. The following cases were run:
5. Borated water injected to lower plenum with a flow of 300 kg/s. Start from 0 kg/s, then at 1 s ramping up reaching 300 kg/s at 2 s, continuing at 300 kg/s for the rest of the run. Coolant flow kept at nominal value of 14500 kg/s.
6. Coolant flow reduced to 1000 kg/s. Start from 14500 kg/s, then at 1 s ramping down to 1000 kg/s at 6 s, continuing at 1000 kg/s for the rest of the run.
2.6.2 A Realistic ATWS Simulation
To test the performance outside synthetic, simplified test cases, a test run on the full ABWR ATWS was performed. This is the model from which the reduced, simplified model was taken for test runs previously mentioned. Two transients were run, one with the standard model and one with the heterogeneous model enabled.
The simulations are run from the same steady state as in the reduced model. The purpose of run is to observe changes in stability and convergence in of the model, and to see if the contribution changes values relevant to acceptance criteria.
The ATWS transient is a case where the reactions are to be halted, but where
the control rods do not deploy. In such an event a predefined routine is run, where
circulation flow is reduced and boron is automatically injected into the nuclear core.
Chapter 3
Results
HetXS refers to the heterogeneous model in the figure legends below.
3.1 Transient Simulations
Below follows results of the transient simulations, presented case-wise.
25
Figure 3.1: Case 1, 300 kg/s boron injected, k
effvs time.
Figure 3.2: Case 1, 300 kg/s boron injected, boron concentration & concentration
difference vs. time, from central active bundle and bypass.
3.1. TRANSIENT SIMULATIONS 27
Case 2
Figure 3.3: Case 2, circulation flow reduced to 1000 kg/s, k
effvs time.
Figure 3.4: Case 2, circulation flow reduced to 1000 kg/s, bypass density difference
from reference vs. time.
Figure 3.5: Case 3, 7 kg/s boron injected, circulation flow reduced to 1000 kg/s, k
effvs time.
Figure 3.6: Case 3, 7 kg/s boron injected, circulation flow reduced to 1000 kg/s,
boron concentration & concentration difference vs. time, from central active bundle
and bypass.
3.1. TRANSIENT SIMULATIONS 29
Figure 3.7: Case 3, 7 kg/s boron injected, circulation flow reduced to 1000 kg/s, bypass density difference from reference vs. time.
Case 4
Figure 3.8: Case 4, 300 kg/s boron slowly injected, k
effvs time. Long runtime.
Figure 3.9: Case 4, 300 kg/s boron slowly injected, average total boron concentra- tion vs. time. Long runtime.
3.1.2 Multiple Bypass Case 5
Figure 3.10: Case 1 & 5, 300 kg/s boron injected, k
effvs. time. Multiple bypass.
3.1. TRANSIENT SIMULATIONS 31
Figure 3.11: Case 5, 300 kg/s boron injected, (C
B)
bpvs. time. 2-channel bypass.
Figure 3.12: Case 5, 300 kg/s boron injected, (C
B)
bpvs. time. 2-channel equal size
bypass.
Figure 3.13: Case 5, 300 kg/s boron injected, (C
B)
bpvs. time. 5-channel bypass.
Case 6
Figure 3.14: Case 2 & 6,circulation flow reduced to 1000 kg/s bypass density dif-
ference from reference vs. time. Multiple bypass.
3.1. TRANSIENT SIMULATIONS 33
Figure 3.15: Case 6, circulation flow reduced to 1000 kg/s, ρ
bpvs. time. 2-channel bypass.
Figure 3.16: Case 6, circulation flow reduced to 1000 kg/s, ρ
bpvs. time. 2-channel
equal size bypass.
Figure 3.17: Case 6, circulation flow reduced to 1000 kg/s, ρ
bpvs. time. 5-channel bypass.
3.2 Full Scale ABWR ATWS
Figure 3.18: ATWS power comparison. Red is with the heterogeneous model, green
without.
3.2. FULL SCALE ABWR ATWS 35
Figure 3.19: ATWS k
effcomparison. Red is with the heterogeneous model, green without.
Figure 3.20: ATWS cladding temperature comparison. Red is with the heteroge-
neous model, green without.
Figure 3.21: ATWS containment temperature rise comparison. Red is with the heterogeneous model, green without.
Figure 3.22: ATWS boron distribution vs. time. With heterogeneous model.
Chapter 4
Discussion of Results
The results obtained are in the form of plots, presented in chapter 3. The system complexity makes any theoretical prediction of significance impossible.
4.1 General
To verify computations one usually refers to an equivalent computation performed, either by simplified theoretical arguments, or by simulations or with simulation code known to be accurate. The problem with verifying transients is that a reference so- lution seldom exists; which is the case for the transients studied in this report, with no other transient simulation software available for comparison.
The fact that the heterogeneous model should have a noticeable effect on k
effcan be visualized by comparing the magnitude of the macroscopic cross section contri- butions of boron absorption of thermal neutrons for base conditions and the hetero- geneous contribution. By choosing a specific volume cell, record the state (density, density history and burnup) and then compare the magnitude of the heterogeneous boron contribution with the standard model boron contribution, one can see under what conditions the heterogeneous contribution becomes significant. At any such time there should be a noticeable difference in k
eff.
The boron coefficients for generating macroscopic thermal absorption cross sections for a typical
1, core node are:
BHSGA2 = 0,683 · 10
−5, BSIGA2 = 1,01 · 10
−51Typical for the simulations performed, with SVEA-96 Optima2 fuel, with burnup of 40 MWd/kgU, historic- and current density of 460 kg/m3.
37
than in the active channel the boron cross section will get a positive contribution, increasing the effect of boron as compared to the standard homogeneous treatment.
On the other hand if there is a higher concentration of boron in the active channel, the opposite will happen and boron efficiency will decrease.
One basic assumption that is made in the modeling of boron transport in POLCA-T is that physical diffusion and turbulent mixing of boron is negligible. Boron trans- port is driven by convective flow alone. In addition, the model assumes that boron is present only in the liquid phase as the vapor pressure of the boron compounds is too low to expect any boron to be present in the steam. Thus, the flow of boron is treated as a slug flow, being carried with the liquid. The liquid flow velocity may very well be different inside the active channel than the water in the bypass re- gion, and this will produce concentration differences between the active and bypass channels. The flow velocity is determined by a numbers of factors such as channel pressure drop, two-phase flow, coolant density, channel orificing, which all have an effect on liquid flow velocity. In the current test model, the flow velocity in the bypass, at steady-state, is lower than in the active channel. Smaller openings in the bypass channel produces higher pressure drops and lower flow velocities, and thus the boron enters the bypass channel at a slower rate.
4.2 Single bypass
In figures 3.1 & 3.2, depicting case 1, the boron enters through the active channels
first, since the boron is injected from below in the reduced model used for verifi-
cation purposes, and since the active channels draws more coolant from the lower
plenum than the bypass, due to higher flow velocities in the active channels. This
means that on average, less boron will be present in each fuel assembly (active and
bypass together) than what is assumed in the standard method (where it is assumed
that the same concentration is present in both active and bypass). With heteroge-
neous model, this concentration difference will change the total macroscopic cross
4.2. SINGLE BYPASS 39
section; in this case lowered, since there is less boron in the assembly than the assumption made in the standard method. This leads to a situation where boron efficiency is reduced with the result that the reactivity starts to increase after an initial decrease. The largest positive reactivity insertion rate occurs when the con- centration difference is at its largest value, which is in accordance with what is expected from observation in equation 2.4 When there is more boron in the active channel than in the bypass, a positive correction term is added to k
eff. Once the ini- tial spatial distribution has settled and a homogeneous state has been reached, k
effof the heterogeneous model model falls and levels out close to the standard model’s steady state k
eff. These results indicate that the effects of the boron contribution of the heterogeneous model model are temporal and that the boron contribution from heterogeneous model vanishes in steady state conditions.
The effects of a reduction in core coolant re-circulation flow are illustrated in figures 3.3 & 3.4. The effective coolant density model and the heterogeneous model model are almost identical and the k
effof the standard model lies slightly lower than the other models down at the minimum, but when k
effincreases, the standard model lies above the effective model and heterogeneous model. This behavior can be explained by the fact that both heterogeneous model and the effective void model are subject to a negative reactivity contribution from the bypass void. From figure 16 one can conclude that early in the transient the heterogeneous model and effective void models will experience a positive reactivity feedback. After about 8 s, this quantity changes sign and starts to give a negative contribution instead. These effects are also seen in figure 3.5 where the Standard model (neutronically unaware of bypass density) has lower reactivity than heterogeneous model and effective void, but once the bypass density falls below the reference density the standard model has a higher reactivity.
For case 3 with disturbance in both boron and lowered circulation flow, seen in figures 3.5 to 3.7, the combination of the results previously discussed can be noticed.
The explanation for the large heterogeneity seen in boron is that the active channel 202 is used for comparison, and not the general boron levels of the active channels vs. the bypass. The active channel 202 has a central placement and thus is strongly affected by the lowered circulation flow. The fact that k
efffor the effective model and the heterogeneous model converge at the end of the figure is caused by the heterogeneities in boron and void act in opposite directions and are of approximately equal size, such that the heterogeneous terms cancel.
To demonstrate that a heterogeneous boron distribution is a transient effect, one
long transient was run with a slowly increasing boron concentration, seen in figures
3.8 & 3.9. As can be seen, heterogeneous model then behaves much like the standard
model, confirming that in slowly evolving transients the boron distribution is much
more uniform.
to better simulation result.
It would seem that by a finer partitioning, the crest seen in the run with a single bypass decreases in size as well as the transient effect being more stretched in time.
Looking at figures 3.11 through 3.13, depicting the averaged bypass boron concen- trations for the different multiple bypass mappings together with the average bypass boron concentration of the single bypass mapping, it can easily be seen that the two channel mapping, mapped after the throttle zones, follows the concentration of the single bypass mapping closely. This together with the fact that higher boron concentrations reach the more active channels with a split bypass channel (about the same average concentration as a single channel, but distributed with higher concentrations at the active parts) leads to an even more heterogeneous effect on k
eff, something that can be seen in figure 3.10. The boron concentrations for the other mappings seem to differ slightly from the single bypass case, with the 5 bypass channel mapping increasing slightly faster than the single bypass early on in the transient, and then slower in the second half, as seen in figure 3.13. The average boron concentration of the 2 bypass channels with equal flow areas is slower as well, which can be seen in figure 3.12. Since it takes longer for boron to reach every- where in the core with these mappings, one understands why in figure 3.10, these two mappings have smoother reactivity curves.
For the disturbance in coolant flow, seen in figure 3.14 through 3.17, the effect of the bypass mappings is not so prominent, probably due to non-radial dependence of void levels. Another contributing factor might be that the multiple bypasses are connected through the upper plenum, which in turn is connected to the uppermost cells of all bypass channels; the very position where the void is spreading from in case the bypass voids (as have been mentioned earlier, the bypass channels are not connected to each other anywhere but at the upper and lower plenums). The different mappings follow the single bypass solution approximately, with 5 bypass channels slightly below during one time window, and then above in a different one.
The deviations are relatively small, about 5 % around the values for the single by-
pass.
4.4. REALISTIC ATWS SIMULATION 41
Comparing average bypass densities with the result from the single bypass chan- nel, 3.16 & 3.17 indicates that the 2 bypass channel mapping with equal flow area and 5 bypass channel mapping, seems to be averaging the densities better, giving smoother curves. Interesting is that in figure 3.17, the average density for the 5 channel mapping passes right through the small nick in the curve of the single by- pass transient. This would seem to show that the 5 channel mapping follows the dynamics of the single bypass well, but is not as sensitive to changes in density. A possible answer to this behavior could be that several channels are more adaptable so that each individual channel can more easily follow changes in the local environ- ment, allowing the individual channels to more easily absorb local fluctuations and disturbances, keeping them local and thereby not spreading to, and affecting the whole system. This would in turn give a smoother behavior of the system. Seeing how the bypass represents a large amount of water, rapid fluctuations needs a lot of force to quickly change the density state of the bypass, thus a smoother behavior would seem more physically sound. With such complex models as those present in POLCA-T, it is however difficult to draw precise conclusions of how interactions take part.
4.4 Realistic ATWS Simulation
Looking at figure 3.19 one can observe a difference between the standard model and the heterogeneous model model; k
effdecreases quicker for heterogeneous model than the standard model. Figure 3.21 shows a result of importance: the heat released to the containment by heterogeneous model model is slightly lower than with the standard model. The reason for this is that the power generated with the heterogeneous model model is slightly lower than the standard in the region of about 220-300 s. This difference can be explained by a heterogeneity in the boron concentration, seen in figure 3.22, starting at 220 s which makes the heterogeneous model model give negative feedback to reactivity.
Apart from differences mentioned above, the results are similar to the standard cross section model. Other state parameters differ somewhat, but this could very well be from feedback from the previously mentioned effects.
4.5 Uncertainties in Results
All results presented are output from a simulator, which is susceptible to numerical
errors, discretization errors in both time and space, e.t.c. Each new addition to
the code has to pass through rigorous verification and validation, and if possible,
comparison to real world measurements. For the simulations run in this thesis,
focused on properties in the context of injection of borated water, there are no real
world data. Boron pollutes the nuclear core significantly and any such experiment
heterogeneity model accuracy). Since the synthetic transients in case 1-3 has a pos- itive heterogeneous contribution to k
eff, these results are conservative. In the full ABWR ATWS simulations however, the results are not conservative, but will have non-conservative errors in the total k
effof about 100 pcm on average. The standard model is not conservative in under these conditions neither. In fact, according to the internal Westinghouse report, the instances when the heterogeneous model has non-conservative errors, the standard model is usually also non-conservative, but with larger errors.
As was mentioned in 4.3, the results of multiple bypasses are not verified and thus
the results should not be trusted too much for analytical purposes.
4.5. UNCERTAINTIES IN RESULTS 43